13.07.2015 Views

View - Kowalewski, M. - Virginia Tech

View - Kowalewski, M. - Virginia Tech

View - Kowalewski, M. - Virginia Tech

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

PALEONTOLOGICAL SOCIETY PAPERS, V. 8, 2002Alexander, pers. comm., 2002] can easily beremoved by taphonomic processes).The evaluation of paleoecological data onpredation can involve both experimental andindirect taphonomic approaches. An experimentalapproach is exemplified by the work of Kaplan andBaumiller (2000), who performed a series ofexperiments with casts mimicking the morphologyof the studied prey (the brachiopod Onniella) andused the results of their experiment to evaluate fortaphonomic biases.Taphonomic data can also be used indirectlyto evaluate biases. Roy et al. (1994) suggested aset of simple questions to assess for taphonomicbias in drill holes. Do fragments of prey skeletonfrequently show partial traces? Frequent partialtraces indicate that most of the fractures passthrough the trace fossils and may have beeninitiated by those traces. This pattern suggests thata bias exists. Are fragments generally common inthe fossil assemblage? If they are rare, thefragmentation bias cannot be severe.Nebelsick and <strong>Kowalewski</strong> (1999), in a studyof drilling predation on echinoids, proposed a simpletaphonomic model to test for bias. They argued that,if drill holes have no taphonomic effect, theirdistribution should be independent of thetaphonomic alteration of drilled tests (i.e., uniformacross taphonomic grades); if drill holes affect thepreservation potential of echinoids by substantiallyweakening their tests, the proportion of drilled testsshould decrease with the increase in taphonomicalteration; and, finally, if drill holes are oftaphonomic (post-mortem) origin (i.e., they arepseudo-drillings), a proportion of drilled tests shouldincrease with the increase in taphonomic alteration.Nebelsick and <strong>Kowalewski</strong> (1999) then showed thatdrilled specimens are as common among testsseverely altered by taphonomic processes as amongtests that are still pristine; that is, the proportion ofdrilled tests does not decrease or increase with theincrease in taphonomic alteration of the tests (Fig. 7).They concluded that drill holes are unlikely to havea serious taphonomic effect even for the small, thintests of the clypeasteroid echinoids used in theirstudy. They noted, however, that the neontologicalmaterial they studied, unlike most of the fossilizedtests, was unaffected by compaction, during whichpreferential breakage of drilled tests would be morelikely to occur. Finally, in a daring study, Kaplanand Baumiller (2000) proposed the use oftaphonomic data to correct for biases in aquantitative way. This approach, specificallydesigned for bivalved organisms, estimates themagnitude of the differential bias in the preservationof opposite valves of an organism and uses theresulting estimate to correct the frequency data oftrace fossils found in those valves.METHODOLOGICALRECOMMENDATIONSIt is clear from the above review that there isa multitude of approaches for collecting traces ofpredation, processing and tabulating the resultingmaterials, analyzing the resulting data, andinterpreting the analytical outputs. It is also clearthat interpretations are rarely unambiguousbecause of the complexity of ecologicalinteractions, the confounding effects of abioticfactors, and the obscuring and biasing effects oftaphonomic processes.It would be foolish to suggest at this point (orperhaps at any point in the development of a scientificdiscipline) that we should erect strict guidelinesregarding how to collect, analyze, or interpret the data(see also Feyerabend, 1978). Consequently, whereasI do propose here some general methodologicalrecommendations, they are primarily geared towardmaking our data more compatible and readable (i.e.,more useful to other researchers).1. Given a wide range of data collecting andprocessing strategies, a method that maximizes thecompatibility of resulting data with futurecomparative analyses should be preferred. Forexample, if possible, fine mesh size should be usedin sieving the samples because that way data canbe compared (by eliminating analytically smallspecimens) to other datasets that were processedwith coarser meshes.2. Results should be reported in a clear mannerso that future researchers can combine the reported28

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!